T357109 parsing improvements

README.md

@@ -52,3 +52,11 @@ Install the required PHP via composer:
 ```sh
 composer install
 ```
+
+### Tests
+
+Run the ShEx parser tests with Node.js:
+
+```sh
+node --test tests/shex-parser.test.js
+```

public_html/vue_components/shex-page.html

@@ -65,6 +65,29 @@ <h3 tt='last_items_created'></h3>
 <script>
 'use strict';
 
+function getDirectPropertyPrefixes( schemaText, directPropertyNamespace ) {
+    const prefixRegex = /^\s*PREFIX\s+([A-Za-z_][\w.-]*):\s*<([^>]+)>/gim;
+    const prefixes = new Set();
+    let match;
+
+    while ( ( match = prefixRegex.exec( schemaText ) ) !== null ) {
+        if ( match[2] === directPropertyNamespace ) {
+            prefixes.add( match[1] );
+        }
+    }
+
+    return prefixes;
+}
+
+function getDirectProperty( part, prefixes ) {
+    const match = part.match( /^([A-Za-z_][\w.-]*):(P\d+)\s*(.*)$/ );
+    if ( match === null || !prefixes.has( match[1] ) ) {
+        return null;
+    }
+
+    return { id: match[2], rest: match[3] };
+}
+
 let ShexPage = Vue.extend ( {
     props : [ 'e' ] ,
     data : function () { return { s:{} , data:{} , lads:["labels","aliases","descriptions"] , can_create_new_item:false , creation_data:{} , warning_messages:[] , created_items:[] , vue_key_num:0 , debugging:false , loading:true , j:{} } } ,
@@ -93,18 +116,24 @@ <h3 tt='last_items_created'></h3>
 
             let s = {labels:{} , props:{}} ;
             let to_load = [] ;
-            const regex1 = RegExp('<([^/<>]+)>(?: EXTRA .*?)?\w*\{(.+?)\}','gi'); // THIS DOES NOT DEAL WITH {num} OR {num,num}!!
+            const directPropertyNamespace = new URL( config.wikibase_api ).origin
+                + '/prop/direct/';
+            const directPropertyPrefixes = getDirectPropertyPrefixes(
+                j.schemaText,
+                directPropertyNamespace
+            );
+            const regex1 = /<([^/<>]+)>(?:\s+EXTRA\s+.*?)?\s*\w*\s*\{(.+?)\}/gi; // THIS DOES NOT DEAL WITH {num} OR {num,num}!!
             while ( (m = regex1.exec(t)) !== null ) {
                 if ( m[1] != start ) continue ;
                 let parts = m[2].split(';') ;
                 $.each ( parts , function ( dummy , part ) {
                     part = $.trim(part) ;
                     if ( part == '' ) return ;
-                    let n = part.match(/^wdt:(P\d+)\s*(.*)$/) ;
+                    let n = getDirectProperty( part, directPropertyPrefixes );
                     if ( n == null ) return ;
 
-                    let prop = n[1] ;
-                    let rest = n[2] ;
+                    let prop = n.id ;
+                    let rest = n.rest ;
                     to_load.push ( prop ) ;
                     //console.log(part,'||',prop,'||',rest);
                     s.props[prop] = { min:0 } ;

tests/shex-parser.test.js

@@ -0,0 +1,100 @@
+'use strict';
+
+const assert = require('node:assert/strict');
+const fs = require('node:fs');
+const path = require('node:path');
+const test = require('node:test');
+const vm = require('node:vm');
+
+const shexPage = fs.readFileSync(
+	path.join(__dirname, '../public_html/vue_components/shex-page.html'),
+	'utf8'
+);
+const regexDeclaration = shexPage.match(/const regex1 = ([^;]+);/);
+
+assert.ok(regexDeclaration, 'ShEx shape regex declaration exists');
+
+function getProductionFunction( name ) {
+	const declaration = shexPage.match(
+		new RegExp( `function ${name}\\([\\s\\S]*?\\n\\}` )
+	);
+
+	assert.ok( declaration, `${name} helper exists` );
+	return vm.runInNewContext( `(${declaration[0]})` );
+}
+
+function getStartShapeBody( schemaText ) {
+	const startMatch = schemaText.match( /\n *start\s*=\s*@<\s*(.+?)\s*>/ );
+	if ( startMatch === null ) {
+		return null;
+	}
+
+	const startShape = startMatch[1];
+	const flattenedText = schemaText
+		.replace( /(?<!\<)#.*(\n|$)/mg, '\n' )
+		.replace( /\n/mg, ' ' );
+	const regex = vm.runInNewContext( regexDeclaration[1] );
+	let match;
+
+	while ( ( match = regex.exec( flattenedText ) ) !== null ) {
+		if ( match[1] === startShape ) {
+			return match[2];
+		}
+	}
+
+	return null;
+}
+
+const prefixAndStart = 'PREFIX wdt: <http://example.test/prop/direct/>\n'
+	+ 'PREFIX wd: <http://example.test/entity/>\n'
+	+ 'start = @<human>\n';
+
+test( 'matches the start shape when whitespace precedes its opening brace', () => {
+	const schema = prefixAndStart + '<human> {\n  wdt:P1 [wd:Q1]\n}';
+	const body = getStartShapeBody( schema );
+
+	assert.ok( body, 'the start shape should be detected' );
+	assert.match( body, /wdt:P1/ );
+} );
+
+test( 'continues to match compact shape syntax', () => {
+	const schema = prefixAndStart + '<human>{\n  wdt:P1 [wd:Q1]\n}';
+	const body = getStartShapeBody( schema );
+
+	assert.ok( body, 'the start shape should be detected' );
+	assert.match( body, /wdt:P1/ );
+} );
+
+test( 'matches shapes with an EXTRA declaration', () => {
+	const schema = prefixAndStart
+		+ '<human> EXTRA wdt:P31 {\n  wdt:P1 [wd:Q1]\n}';
+	const body = getStartShapeBody( schema );
+
+	assert.ok( body, 'the start shape should be detected' );
+	assert.match( body, /wdt:P1/ );
+} );
+
+test( 'allows whitespace around EXTRA and before the opening brace', () => {
+	const schema = prefixAndStart
+		+ '<human>   EXTRA   wdt:P31   {\n  wdt:P1 [wd:Q1]\n}';
+	const body = getStartShapeBody( schema );
+
+	assert.ok( body, 'the start shape should be detected' );
+	assert.match( body, /wdt:P1/ );
+} );
+
+test( 'resolves custom prefixes for the local direct-property namespace', () => {
+	const directPropertyNamespace = 'http://example.test/prop/direct/';
+	const schema = 'PREFIX wdt: <http://example.test/prop/direct/>\n'
+		+ 'PREFIX wfdt: <http://example.test/prop/direct/>\n'
+		+ 'PREFIX wdtForeign: <http://www.wikidata.org/prop/direct/>\n'
+		+ 'PREFIX ex: <http://example.test/prop/>\n';
+	const getDirectPropertyPrefixes = getProductionFunction( 'getDirectPropertyPrefixes' );
+	const getDirectProperty = getProductionFunction( 'getDirectProperty' );
+	const prefixes = getDirectPropertyPrefixes( schema, directPropertyNamespace );
+
+	assert.equal( getDirectProperty( 'wfdt:P1 [wd:Q1]', prefixes ).id, 'P1' );
+	assert.equal( getDirectProperty( 'wdt:P2', prefixes ).id, 'P2' );
+	assert.equal( getDirectProperty( 'wdtForeign:P3', prefixes ), null );
+	assert.equal( getDirectProperty( 'ex:P4', prefixes ), null );
+} );